Hello, Ketan,

Inline...

On 7/9/19 08:14, Ketan Talaulikar (ketant) wrote:
> < An engineer comes out from the trenches/ >
> 
> Hi Fernando,
> 
> I will attempt to answer and but also setup the context first.

Thanks for elaborating, by the way.


> 4) This is not about the IPv6 Internet. All of SR is about within a SR domain 
> and as an extension so is SRv6 and SRH.

At least at the time of this writing, we don't have "flavours" of
IP/IPv6. The same protocol is employed everywhere, because that's the
scope of the layer.




> There are exactly two "features" (at least AFAIK) of a large set where we 
> encounter scenarios where SRH insertion/removal for this outer IPv6 header 
> when the customer packet transits the SR domain:
> a) BSID (End.B6.Insert) : this is a SR Policy anywhere in the transit path 
> for TE and/or Service Chaining purposes
> b) TI-LFA/Microloop Avoidance (T.insert) : These are protection techniques 
> for sub-50msec fast-reroute and repair for failures in this transit network
> 
> You can get more details/references on them on this thread : 
> https://mailarchive.ietf.org/arch/msg/spring/Jtg9YPxEjes4Bm-1ngwXmJfeybs 
> 
[....]
> 
> These features exist and are widely deployed today with SR-MPLS and work 
> similarly - difference is working with stacks of label instead of SRH.
> 
> Now, it is most important to understand that the encap variants of these 
> features (End.B6.Encap & T.encap) are already defined in the SRv6 spec.
> 
> So then we come to your Q - why the insert flavors? The motivation is to 
> reduce the overhead of *yet another* IPv6 header and encap when doing these 
> features. You will also realize by now that this aspect is not specific to 
> SRv6/SRH but in general to any solution based on IPv6. Anyone having some 
> decent experience with SR will vouch for the importance of these two features.

So, double-checking if I understood correctly: You are saying that the
two uses cases that you are referring to already have an alternative
specification with encap/decap, but this document proposes to use EH
insertion to avoid the extra overhead of adding an additional IPv6 header?

Thanks!

Cheers,
-- 
Fernando Gont
SI6 Networks
e-mail: fg...@si6networks.com
PGP Fingerprint: 6666 31C6 D484 63B2 8FB1 E3C4 AE25 0D55 1D4E 7492




_______________________________________________
spring mailing list
spring@ietf.org
https://www.ietf.org/mailman/listinfo/spring

Reply via email to